White: Glawurtz89 - all-play-all tournament, ChessWorld.net, 2018
Twenty years ago in an ICCF thematic I played the novelty 19 Rf5 h6 20 Bf4 h5!. This was my improvement on analysis (20...Rfe8 21 Rg1) by Victor Ivanov and Alexander Kulagin in their very interesting book Play the Schliemann Defence! (Olbrich 1994).
It was indeed good, but I failed immediately to back it up. After 21 Bg5, my next move 21...Kc7? could have been met by 22 Qd2!, since 22...hxg4? 23 Qa5+ Kc8 24 Qxa7 g3! (as I'd analysed) fails instead to 24 Nc4! and if 24...g3 then 25 Nxd6+ Rxd6 26 Bf4 Rd7 27 Qxa7 wins for White. Fortunately, my opponent played 22 Ng6?, made some more mistakes, and finally allowed me a win with 31...f2! 32 d5 R8e4! 33 dxc6+ Kc7 – which I missed. I didn't even realize I was winning. This is all according to the engine of course.
Going back, the correct move was 21...Bc7, which seems to work out for Black. In the game below I intended to give it a go – until my opponent played 19 Rg1!, planning simply to trap the queen with Rg3.
No worries. I'd looked at this back in the day. Ivanov & Kulagin analysed the further 19...Bxe5 20 dxe5 Rxe5 21 Rg3 Qxg3 22 hxg3 Rxg5 23 Qe3 Rxg4 24 Qxa7 Re8, giving the appropriate symbol for “with counterplay”.
Okay, I'll just check that with the engine... 25 Rd4 Rg5 26 Rb4 b5 27 Rb3 Rh5 28 Ra3 e3 29 Rxe3 “+- (2.22)”. In human-speak: “White is winning”. Bollocks!
I had to leave the g4-pawn alone, grovel with 23...Re5 24 Qxa7 Rfe8, and hope that the e-pawn would still offer enough.
Well, as it turned out, my rooks managed to get in on the second (at the cost of a couple of pawns) and eliminate the white queenside, leading to a rook plus one vs. knight plus two endgame where the black e-pawn was enough. For instance, at the end, 46 Rxh7 c4+! 47 Kxc4 e2 48 Rh1 Ne3+ 49 Kd3 Nf1 draws.
All the same, if there are any novelties to be found post 23...Re5, they're going to be White's. I think I'll look for an improvement earlier on next time.
